Sorry Traci -- I'm usually good about including this:

login pub
password download

On 10/08/2009 03:16 PM, Sandoval, Traci I wrote:
> Donna, the link appears to be password protected, thank you Traci 
>
> http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2009_09/
>
> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of Donna Dierker
> Sent: Thursday, October 08, 2009 1:42 PM
> To: Caret, SureFit, and SuMS software users
> Subject: Re: [caret-users] do spherical registration using freesurfer 
> generated surfaces
>
> Hi Traci,
>
> No, I don't. *Someone* -- YOU KNOW WHO YOU ARE -- has me working on
> similar scripts, but different enough to be both a fair amount of work
> and not as applicable to what I had in mind. I hope to wrap that up this
> week (but we'll see).
>
> Meanwhile, I posted interim versions of what I do have:
>
> http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2009_09/
>
> freesurfer2caret.Ver2a.preborder.mni2wustl.sh - imports Freesurfer to
> Caret and creates inputs for next script
> draw_core6_landmarks.tim.sh - draws borders
> border_variability.sh - generates quantitative measures of landmark quality
> generate_scene.check.sh - generates captures, scenes of borders
> [manual step: Layers: Borders: Border Update]
> postborder.sh - smooths medial wall, morphs and aligns sphere, registers
>
> If you have more than a handful of subjects, it will probably be worth
> getting these to work.
>
> Realistically, I hope to have something applicable to more typical
> Freesurfer input by 10/31.
>
> The script above had more steps like this than the typical stream would
> have, due to strangeness of the input data and inadequate talairach.xfm
> (not imgreg's fault):
>
> caret_command -surface-apply-transformation-matrix
> $CASE.$HEM_FLAG.MIDTHICKNESS.coord $CASE.$HEM_FLAG.topo
> $CASE.$HEM_FLAG.MIDTHICKNESS_orig.coord -matrix $MATRIX
>
> Donna
>
>
> On 10/08/2009 01:15 PM, Sandoval, Traci I wrote:
>   
>> Hi Donna, do you have these cool caret scripts ready ie...
>>
>> Next week, I'll adapt these import scripts to a dataset more like what yours 
>> is likely to be. When I am done, I can post the resulting scripts.
>>   
>>     
>>> Can you ping me in two weeks?
>>>
>>> Donna
>>>     
>>>       
>> Thanks for all your work and help. Traci
>>
>> Traci Sandoval
>> Research Assistant
>> NeuroPsychometric Research Lab
>> School of Behavioral and Brain Sciences - University of Texas at Dallas 
>> Center for BrainHealth - University of Texas at Dallas
>> Department of Psychiatry - University of Texas Southwestern Medical
>> www.utdallas.edu/research/nprlab
>> 972-883-3270
>>
>>
>>
>> -----Original Message-----
>> From: [email protected] 
>> [mailto:[email protected]] On Behalf Of Donna Dierker
>> Sent: Friday, September 18, 2009 11:20 AM
>> To: Caret, SureFit, and SuMS software users
>> Subject: Re: [caret-users] do spherical registration using freesurfer 
>> generated surfaces
>>
>> Traci,
>>
>> Yes, one lab here at Wash U is doing that. A few comments:
>>
>> * Since you don't have a flat map, the reference distance for the
>> posterior sylvian starting point changes from 9mm posterior to the
>> SF/SF2 intersection (relative to the flat map) to 12mm posterior to the
>> SF/SF2 intersection on the spherical map.
>>
>> * It's tougher to discern the full sylvian trajectory on the spherical
>> map, as well as center the sylvian on the main window, such that you can
>> see the whole length. Fortunately, the absolutely fun-to-use Layers:
>> Borders: Border Update makes this easy to fix, if you can't draw the
>> whole length in one go. (If your older Caret version doesn't have this
>> feature, you really MUST upgrade.)
>>
>> * You have to draw the medial wall dorsal and ventral in separate
>> segments, and you have to pay attention to their orientation (dorsal
>> starts anteriorly, ventral posteriorly). Calcarine starts posterior.
>> Flattening took care of this for you.
>>
>> * The lack of a compressed medial wall view makes it harder to draw the
>> medial wall ventral.
>>
>> Donna
>>
>> On 09/18/2009 10:19 AM, Sandoval, Traci I wrote:
>>   
>>     
>>> Donna, That’s sounds great, I have scripts that take my freesurfer brains 
>>> all the way to flattening in caret, but I flatten and draw borders 
>>> manually. I noticed you mentioned you guys are doing the auto-landmarks on 
>>> the spherical surface, and not flattening at all. Can I do that manually as 
>>> well? And skip flattening? Thanks Traci
>>>
>>> -----Original Message-----
>>> From: [email protected] 
>>> [mailto:[email protected]] On Behalf Of Donna Dierker
>>> Sent: Friday, September 18, 2009 9:26 AM
>>> To: Caret, SureFit, and SuMS software users
>>> Subject: Re: [caret-users] do spherical registration using freesurfer 
>>> generated surfaces
>>>
>>> Traci,
>>>
>>> I do have some scripts I could make available now, but in two-four
>>> weeks, I'll have more general purpose ones. The most complete set were
>>> actually developed for a data set segmented in Caret. But they include
>>> the auto-landmarks, border variability (QA), generating scenes/captures
>>> showing the borders, and running registration via command line. There
>>> also are several versions of a "preborder" script that covers importing
>>> from Freesurfer, but most of these used anatomical data that was in
>>> 711-2* space before feeding to Freesurfer. Your data is unlikely to fit
>>> that category.
>>>
>>> Next week, I'll adapt these import scripts to a dataset more like what
>>> yours is likely to be. When I am done, I can post the resulting scripts.
>>>
>>> Can you ping me in two weeks?
>>>
>>> Donna
>>>
>>> On 09/17/2009 04:07 PM, Sandoval, Traci I wrote:
>>>   
>>>     
>>>       
>>>> Donna, I would love to use these scripts that aren't publicized yet. How 
>>>> can I attain said scripts?!! Thank you for all your work and questions 
>>>> answered. Traci
>>>>
>>>>
>>>> Traci Sandoval
>>>> Research Assistant
>>>> NeuroPsychometric Research Lab
>>>> School of Behavioral and Brain Sciences - University of Texas at Dallas 
>>>> Center for BrainHealth - University of Texas at Dallas
>>>> Department of Psychiatry - University of Texas Southwestern Medical
>>>> www.utdallas.edu/research/nprlab
>>>> 972-883-3270
>>>>
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: [email protected] 
>>>> [mailto:[email protected]] On Behalf Of Donna Dierker
>>>> Sent: Thursday, September 17, 2009 8:44 AM
>>>> To: Caret, SureFit, and SuMS software users
>>>> Subject: Re: [caret-users] do spherical registration using freesurfer 
>>>> generated surfaces
>>>>
>>>> Hi Jidan,
>>>>
>>>> See inline replies below.
>>>>
>>>> Donna
>>>>
>>>> On 09/17/2009 02:06 AM, z丹丹 wrote:
>>>>   
>>>>     
>>>>       
>>>>         
>>>>> Hi,
>>>>>
>>>>> I already have fiducial surfaces generated by Freesurfer, which are
>>>>> all in the MNI space. I want to do spherical registration to an atlas
>>>>> spherical surface. The steps i need to do are:
>>>>>
>>>>> 1, generate the ellipsoid, spherical surfaces, and flat maps.
>>>>> 2, create the 6 landmarks on both individual and atlas surface.
>>>>> 3, do spherical mapping.
>>>>>
>>>>> Are the procedure right? Thanks! 
>>>>>     
>>>>>       
>>>>>         
>>>>>           
>>>> Funny you should ask. We (the Van Essen Lab) are using Freesurfer for no
>>>> less than five projects, and we have shell scripts that streamline much
>>>> of the import, border drawing, QA, and registration. There is a feature
>>>> we call "auto-landmarks" that draws the borders reasonably well. Many
>>>> borders need touch-ups, but a single rater fixed 72 hemispheres' borders
>>>> in 1.5 days. If you have only a handful of subjects, it's not clear
>>>> whether it would be worth it for you to use our methods. But if you have
>>>> dozens, then they would save you a lot of time.
>>>>
>>>> We haven't broadly publicized these features/scripts, because they're
>>>> still in the final stages of development and testing. But they have been
>>>> working well for us.
>>>>
>>>> Even if you don't use auto landmarks or these scripts, one question
>>>> stands out in your steps above: Do you need a flat map? No, not really.
>>>> It certainly is easier to draw landmarks on the flat map than the
>>>> sphere; however, I'm not sure it's worth the hassle. Then again, if
>>>> you're not using the scripts, it might be the shortest path to 
>>>> registration.
>>>>   
>>>>     
>>>>       
>>>>         
>>>>> In fact before I move to the 2nd step, I was stuck in the flat map
>>>>> generation. I met two problems:
>>>>> 1, As my surface is in MNI space, the surface orientation is not right
>>>>> in the CARET software. I found that the anterior and posterior are
>>>>> reversed with superior and inferior. Is there any method to roate it
>>>>> into the right orientation in CARET?
>>>>>     
>>>>>       
>>>>>         
>>>>>           
>>>> Yes, but this is not consistent with my experience of MNI space nor
>>>> Freesurfer surface orientation. While MNI volumes are sometimes stored
>>>> using left-handed coordinate systems (what some call radiological
>>>> orientation), the coordinates are always reported with -x being left, +x
>>>> being right; -y being posterior, +y being anterior; -z being inferior,
>>>> +z being superior. I've never seen a Freesurfer surface with the Y and Z
>>>> axes flipped -- not even ones written in MNI space.
>>>>
>>>> It's easy to flip a surface from LAS to LPI
>>>> (http://brainmap.wustl.edu/OLD/SureFit/orient.html) like this:
>>>>
>>>> caret_command -surface-apply-transformation-matrix $COORD $TOPO
>>>> $FLIPPED_COORD -matrix "1 0 0 0 0 -1 0 0 0 0 -1 0 0 0 0 1"
>>>>
>>>> But I really would not brush this off. I would get to the bottom of why
>>>> the supposedly MNI surface is not in the expected orientation. The
>>>> evidence suggests it is not in MNI space, as asserted.
>>>>   
>>>>     
>>>>       
>>>>         
>>>>> 2, The other problem is , I found that the origin of my surface is not
>>>>> AC. So when I flatten the surface, it can't find the correct medial
>>>>> wall to cut. Do you think there is a way to solve this too?
>>>>>     
>>>>>       
>>>>>         
>>>>>           
>>>> This is more evidence that the surface is not really in MNI space. How
>>>> did it get transformed to MNI?
>>>>
>>>> Note that the MNI origin is not exactly the AC; see
>>>> http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach for details. But
>>>> it is certainly close enough to the AC that the compressed medial wall
>>>> view should be centered over the medial wall. The fact that it is not
>>>> indicates a problem with your transformation to MNI space.
>>>>   
>>>>     
>>>>       
>>>>         
>>>>> Thanks a lot!
>>>>>
>>>>> Jidan
>>>>>     
>>>>>         
>>>>>           

_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users

Reply via email to